Tobacco use, primarily cigarette smoking, is the greatest source of preventable mortality in the world and costs over $160 million in health-related economic losses in the U.S. alone. Nicotine dependence is the primary reason that smokers continue smoking and that most unassisted quit attempts fail within a single week. It is known that nicotine dependence has a genetic component, but that it is a complex trait, i.e., no single gene is responsible for nicotine dependence. Thus, researchers and funding agencies have devoted considerable effort and support to identifying the genetic underpinnings of the trait through whole genome scans, putting us in a unique position to identify global genetic predictors of nicotine dependence. This study proposes to realize the promise of the NIDA-funded Collaborative Genetic Study of Nicotine Dependence (COGEND) whole genome data through the accomplishment of two specific aims: (1) to identify the set of genetic variations underlying the complex trait of nicotine dependence using a cutting-edge computational method called Bayesian networks and (2) to validate the prognostic model in an entirely independent population. This proposal represents the very first step of a broader research program aimed at discovering the complex network of interactions underpinning nicotine dependence. The ultimate result of this program will provide a clinical tool, which will accurately assess the risk of dependency, allow for individualized preventive measures, elucidate the molecular processes of dependence and nominate novel targets for the pharmaceutical treatment of nicotine addiction. Nicotine dependence places an enormous burden on individuals and society. Genetic factors are responsible for at least some part of the condition, and the NIDA has already funded a study, called COGEND, that examined over 40,000 genetic variations in people who were nicotine dependent and who were not nicotine dependent. We propose to use cutting-edge techniques to analyze this large dataset to identify a valid predictive model of nicotine dependence that will help us predict, diagnose, and treat this condition. [unreadable] [unreadable]